@Teknium Will eagerly wait to use this once the required PRs have been merged in! (I'd rather just stick to official updates rather than apply random PRs on my own instance...)
English
Sam Wolfstone
663 posts
Sam Wolfstone
@SamWolfstone
Sculpting, AI, Philosophy, Coding
Entrou em Kasım 2020
134 Seguindo246 Seguidores
If you liked lossless claw, here's a new memory plugin that brings it to Hermes Agent 👇
Stephen Schoettler@StephenSch44748
I built an LCM plugin for @NousResearch Hermes Agent. Required a patch to make the context engine pluggable. Lossless Context Management — every message preserved, nothing lost to compaction. PR: github.com/NousResearch/h… Plugin: github.com/stephenschoett…
English
@AcerFur @synthwavedd Such a cool frickin' benchmark!! Very nice work. Can't wait to see how the new GPT one does on this. Do you have other tests you're keeping in your back pocket in case the labs target these specific tests?
English
I'm not one to be too interested in image gen capabilities, but I do care about reasoning capabilities, so I am introducing a new benchmark testing reasoning during image generation.
Introducing the Image Reasoning Generation Benchmark (IRGB):
#irgb" target="_blank" rel="nofollow noopener">pellaml.github.io/iumb/#irgb
English
@synthwavedd @chatgpt21 @DrBeavisAI Can't help but think that 5o and 5.5 will be different models. 5o needs to be really fast/cheap if it's going to power advanced voice mode. If 5.5 is a step change in intelligence, probably huge, slow and expensive.
English
@chatgpt21 @DrBeavisAI I mean naturally they would say that
Internally it's currently slated as 5.5, but 5o isn't out of contention either. Doubt it's 6
English
@johnennis Definitely did at first. Got one of my smartest friends into LLMs so now I have someone who's fairly interested in talking about AI stuff with me. Also have a work-colleague-turned-friend who also loves talking about AI with me now, who has very different views from me
Need more
English
I think one of the biggest challenges when it comes to going hard into using AI is loneliness
I am learning all these awesome things and becoming super capable
But the set of people that I can really talk to about it is very small
Is anyone else having this experience?
English
Sam Wolfstone retweetou
@AcerFur Feel kinda stupid asking but I'm too curious, what makes it not quite pass the test here?
English
maskingtape-alpha
gaffertape-alpha
packingtape-alpha
Seems like good image models on the image arena (try them out)... but they're not quite perfect just yet. Still fails the Rubik's Cube reflection test.
Acer@AcerFur
@m__dehghani Alright, @m__dehghani time for these next: A validly scrambled Rubik's cube placed by a mirror, clearly showing its mirror reflection. No harsh light reflections. Incorrect centres, edge, and corner pairings:
English
@KarolCodes @0xSero @theo Even with some details in the SOUL.md, it's really hard to get GPT-5.4 not to yap. If you ask it to be succinct, it'll just be curt in its sentences but still send you 50 lines in the response...
English
So, uh, what subscription should I be using for my OpenClaw now? 🙃
English
@synthwavedd @sawlygg Was the prompt for this one also fairly simple?
English
yeah it's over holy shit
h/t @sawlygg
leo 🐾@synthwavedd
for those wondering, yes, gpt image 2 (not based on 4o ;3) is now on @arena enjoy
English
@SamWolfstone I have 3 more much larger giveaways in the work, it ain’t easy
English
Do you want 3 months of Codex Pro?
Comment under the post we are selecting a 5 people very soon
3 MONTHS of Pro. I genuinely think this is the highest leverage sub we have
Sarah Chieng@MilksandMatcha
Giving away 5 Codex Pro plans Each person will get 3 months of free Codex Pro (highest tier). Winners will be selected from comments in 48 hours, comment below why you want it.
English
Giving away 5 Codex Pro plans
Each person will get 3 months of free Codex Pro (highest tier).
Winners will be selected from comments in 48 hours, comment below why you want it.
OpenAI@OpenAI
Today, we closed our latest funding round with $122 billion in committed capital at an $852B post-money valuation. The fastest way to expand AI’s benefits is to put useful intelligence in people’s hands early and let access compound globally. This funding gives us resources to lead at scale. openai.com/index/accelera…
English
If we made /slow mode in Codex, would you use it? What for?
(Slower inference at a cheaper cost)
English
Sam Wolfstone retweetou
Claude, in a world full of unknown unknowns: Good, now I have a complete picture
English
@robbensinger @theaidocfilm I wish it was showing somewhere in central London :(
English
I've seen @theaidocfilm three times, I have never been this excited to rewatch a movie?? feed me your reactions if you've seen it
English
@EntropyChase @catehall Also, happy to continue in DMs if you'd like, as tweets are a bit annoying for this type of convo :)
English
@EntropyChase @catehall of things in the world, and they'd still be able to 'understand' those things, even if those things were incorrect. Maybe you and I have a different definition of 'understanding', so maybe we're slightly talking past each other...
English
